Sampling

Sample Size

Focal Research has industry-recognized expertise on sampling and sampling issues. Our experience with longitudinal and tracking studies in particular has allowed for the testing and development of techniques, and has proven the accuracy of results.

We have also specific experience in sampling rare populations, particularly in the social research arena, and have provided international consultation services on methodologies and survey techniques for such groups.

Why is sampling important?

Sampling is a key component of marketing research studies, and substantially impacts the representativeness of the results. In fact, the principal source of error in research is sampling. This is particularly critical in a continuous (on-going) study since differences in results between measurement periods may be due to sampling error rather than actual changes in market response. Therefore, for results to accurately reflect market response, be generalizable to the underlying population and have validity over repeated measurements, considerable effort must be directed towards sampling.

Focal Research rigorously monitors and manages the sample throughout data collection to maximize the response rate achieved. Interviewers are required to make a minimum of seven callbacks to a telephone number on the sample for a random selection sampling methodology. However, it is possible that more than twenty calls will be made as numbers are not dropped from the sample listing during data collection. Calls are placed to eligible numbers as long as data collection continues. All numbers on the sample are tried at various times of the day and evening, and on various days of the week in attempts to screen for qualified respondents. This ensures the maximization of response rates which results in better representation of the underlying sample list. In contrast, many of the current CATI (Computer Automated Telephone Interviewing) systems drop numbers from the sample listing, often after only three calls are made to that number, possibly in the same day/evening. This can artificially inflate response rates causing the sample list to appear to be better represented than it actually is.

The lack of such sample management techniques and standards can result in a convenience or non-probability sample, such as quota sampling, which may be reasonable for exploratory designs in which the emphasis is on generating ideas (i.e., focus groups). However, the problem with convenience samples is there is no way of determining whether it is representative of the underlying or target population, "although there is temptation to conclude that large samples, even though selected conveniently, are representative...which they are not." (Gilbert Churchill, Marketing Research Methodological Foundations, 1987)